In the last two decades, the focus of digital image processing and computer vision research community was on low-level, mid-level feature representation and various hybrid approaches based on feature combination. These approaches have shown good results in various computer vision applications such as image segmentation, character recognition, biomedical image analysis, content-based image retrieval, video analysis, traffic signal recognition, document analysis and recognition and many other relevant fields.

Deep learning is considered an advanced field of machine learning. It requires large-scale data for training. Recently, deep learning models have been applied to various computer vision applications. The performance of deep learning models depends on the amount of training data and on large-scale data. Deep learning models have shown a better performance than the traditional machine learning approaches. However, due to the advancements in deep learning models, traditional computer vision techniques are now obsolete. Factors such as computation time, computation power, accuracy, and number of input parameters are the drawbacks of deep learning models. There are hybrid approaches based on traditional machine learning models that can perform better than deep learning models based on the available samples of training data.

The aim of this Special Issue is to bring together original research and review articles discussing hybrid approaches for image and video processing. We welcome submissions discussing a clear comparison between the low-level feature representation, and the mid-level feature representation. Research should include applications such as image segmentation, image classification, character recognition, and biomedical image processing. Moreover, we also welcome research mentioning applications in video processing, video or image annotation, document image processing and segmentation, document retrieval, symbol recognition, traffic sign, board detection, and action recognition in video. We hope that this Special Issue gathers research demonstrating that traditional approaches and hybrid approaches can dominate deep learning models or vice versa.

Potential topics include but are not limited to the following:

• • Image segmentation and classification
• • Texture analysis
• • Shape-based image representation
• • Character and face recognition
• • Biomedical image analysis
• • High spatial resolution image analysis
• • Video processing or video classification
• • Video or image annotation
• • Content-based image retrieval
• • Document image processing and segmentation
• • Document retrieval
• • Noise reduction algorithms for image classification
• • Symbol, sign board and traffic signal recognition
• • Action recognition in video
• • Transfer learning and data augmentation